Highly-selective separation of divalent ions from seawater and seawater RO retentate

https://doi.org/10.1016/j.seppur.2016.10.030Get rights and content

Highlights

  • A new method is shown for highly selective separation of seawater divalent ions.

  • The 3-step approach comprises of two NF steps followed by six DiaNF cycles.

  • TH:SO42− ratio in DiaNF influent governs Cl concentration in product solution.

  • Product solution characterized by [Mg2+] ∼5000 mg/l, with much lower [Cl] & [Na+]

Abstract

A membrane-based process is presented for separating divalent ions (namely Mg2+, Ca2+ and SO42−) from seawater in a highly selective fashion. The main goal is to selectively and cost-effectively separate Mg2+ from seawater with the intention of either dosing it into desalinated water in the desalination post-treatment stage or generating a Mg2+/Ca2+/SO42− solution that can be used for, e.g., precipitating struvite from high-strength wastewaters prior to anaerobic-digestion, with minimal addition of detrimental Cl/Na+ ions. The process comprises three steps: First, seawater undergoes a high-recovery nanofiltration (NF) step with an “open” NF membrane aimed at reducing the molar ratio between divalent-cations and SO42− in the NF retentate. The retentate of this step is then subjected to further ∼65% recovery (conventional) NF step aimed at increasing the Mg2+ concentration in the retentate, which is thereafter subjected to a diananofiltration step aimed at reducing the monovalent ion concentrations while maintaining high Mg2+/SO42− concentrations in the product solution. The hypothesis (which was fully substantiated in the work) was that the reduction of the molar ratio between total hardness and SO42− in the retentate of the 1st NF step would result in a lower Cl to Mg2+ concentration ratio in the product solution (retentate) of the NF-NF-DiaNF process sequence. Results are presented for a variety of operational conditions using both seawater and seawater reverse osmosis brine as raw solutions. The cost of separating one kg of Mg2+ from seawater using the method is significantly lower than the equivalent cost of the chemical MgSO4 in all the presented scenarios. However, reducing the Cl concentration in the product solution by ∼62% more than doubled the cost. For example, dosage of 20 mg Mg2+/l to desalinated water was estimated at $0.006/m3 and $0.017/m3 for addition of 32 and 12 mg Cl/l along with the Mg2+, respectively.

Introduction

Seawater contains a relatively high concentration of Mg(II) (1350–1500 mg/l), along with a very high concentration of Cl (∼20,000 mg/l) and Na+ (∼11,000 mg/l) and varying concentrations of many other constituents. Multiple processes have been developed over the years to recover Mg2+ from seawater, for use in a variety of applications (e.g. generation of solid Mg [1], [2], [3]; dosing Mg2+ into soft and desalinated waters [4], [5]; producing Mg-rich solution used to enhance struvite precipitation from wastewater streams [6], [7], [8], etc.). The degree of selectivity with which the Mg(II) is extracted from seawater varies considerably between published separation processes and so does the estimated cost for extracting a mass unit of Mg(II). Recently, Nativ et al. [9] proposed an inexpensive process for separating Mg2+ from seawater and seawater reverse osmosis (SWRO) brine, which is based on nanofiltration (NF) followed by diananofiltration (DiaNF). Using the mass ratio between Cl and Mg2+ in the extracted solution as an indicator for selectivity Nativ et al. [9] attained a minimal value of 1.5 (kg Cl) to 1 (kg Mg2+) along with a highly competitive cost ($0.4 per kg Mg). However, in certain cases, in which the addition of Cl ions with the extracted Mg(II) solution is undesirable, a much lower ratio may be required. One such case relates to the use of the Mg-rich solution for enhancing struvite precipitation from ammonia-laden high-strength organic solutions, as pretreatment to anaerobic digestion [7], [8]. In this case the aim of the struvite precipitation step is both to recycle nutrients but also to reduce the negative effect that toxic NH3(aq) has on downstream methanogenic activity. However, examining the inexpensive methods suggested thus far (from the Cl to Mg2+ ratio standpoint) reveals that a large mass of unwanted constituents, including Cl, may be added along with the Mg2+. For example, applying the method suggested in [8] for removing ∼25 mM of NH4+ (∼350 mg N/l) through struvite precipitation, using a DiaNF retentate solution characterized by a Cl:Mg2+ mass ratio of 1.5–1 (the value reported in [9]) would increase the Cl concentration in the anaerobic digestion wastewater by ∼1330 mg Cl/l, which, in all likelihood, would jeopardize the activity of methanogenic bacteria and thereby the anaerobic digester stability. A second example for cases in which the Cl concentration should be minimized relates to the use of the DiaNF Mg-rich retentate for supplying Mg2+ to soft or desalinated waters. In the event that the soft water is used for irrigation purposes (either directly or as treated wastewater) a restriction on the Cl concentration added with the Mg2+ may be in place. In Israel, large desalination plants’ bids require that the water leaving the desalination plant will be characterized by [Cl] < 20 mg/l and [Na+] < 30 mg/l (the higher allowed Na+ concentration takes into account NaOH usage in the desalination post-treatment step). Since the Cl concentration following a two-pass seawater desalination process is typically <10 mg Cl/l, the 20 mg/l requirement translates into maximal additional ∼10 mg/l associated with the Mg2+ dosage, delineating the need for improving the selectiveness of the separation method and generate product solutions with significantly reduced Cl to Mg2+ ratio. The process developed in this work focused on the latter rationalization, i.e. addition of Mg2+ to desalinated waters, in which a restriction exists on the maximal chloride concentration in the product water.

A post treatment (PT) step is practiced in almost all desalination plants for ensuring the stability of the water within the distribution system, its suitability for irrigation and the presence of minerals required for proper humans’ diet. Typical PT processes supply the water with bicarbonate and Ca2+ ions and adjust pH to set a small (typically >3 mg/l as CaCO3) positive calcium carbonate precipitation potential. Lately, however, both the world health organization (WHO) and the Israeli Ministry of Health recommended maintaining a minimum Mg2+ concentration in drinking water (10 and 20 mg/l, respectively) for reasons related to public health [10], [11], [12], [13]. Recognizing that adding Mg2+ ions through dosage of food-grade chemicals (e.g. MgSO4) is relatively expensive (see cost analysis at the end of the paper), development of inexpensive methods for supplying desalinated water with Mg2+ ions while at the same time ensuring minimal addition of unwanted components (e.g. Na+, Br, Cl, and B), is essential.

The process described in this paper utilizes two types of nanofiltration (NF) membranes for selectively separating Mg2+ from seawater and dosing the extracted Mg2+-rich solution to desalinated water. The very high selectiveness for divalent ions demonstrated by the process is attained by applying a DiaNF step. DiaNF involves continuous dosage of low-TDS water to the feed of the NF process (at a flowrate similar to the rate of the permeate flowing out of the NF membrane) to result in removal of charged and uncharged species that permeate preferably through the membrane. When DiaNF is applied on seawater solutions, the species that are washed-out (i.e. which preferably pass the membrane to the permeate side) are mainly mono-valent ions, CO2 and B(OH)3, while the multi-valent ions (namely SO42−, Ca2+ and Mg2+ but also Sr2+) are well rejected and thus accumulate in the retentate. A “cycle” in DiaNF refers to addition of low-TDS water at a volume equal to given volume of the feed water, while producing similar volume of permeate.

Direct DiaNF of seawater was first suggested by Oumar et al. [14] with the purpose of separating divalent from monovalent ions. Nativ et al. [9] extended the concept by applying a two-step process comprising of an NF step followed by DiaNF step. From a process point of view, the purpose of the preceding NF step was to both decrease the ratio between Cl and Mg2+ and to increase the Mg2+ concentration in the retentate on which the DiaNF step was subsequently performed. In this way, the volume of the costly low-TDS solution used in the DiaNF step could be considerably reduced (per unit mass of extracted Mg), lowering the cost of separating one kg of Mg2+ by the method to below ∼$0.4. However, as mentioned above, the lowest [Cl] to [Mg2+] weight ratio that could be attained by this process was ∼1.5–1. Thus, when the product solution of this process would be dosed to desalinated water to attain Mg2+ concentration of 20 mg/l, as required by current Israeli (tentative) regulations, ∼30 mg/l of Cl would also be added. Such Cl addition surpasses the concentration allowed according to Israeli SWRO desalination bids. Nativ et al. [9] reported that the reason for the inability to reduce the Cl concentration to lower values lies in the inherent two-fold difference between the total hardness (TH) concentration (i.e. [Ca2+] + [Mg2+] + [Sr2+], amounting to ∼126 meq/l) and that of SO42− (∼60 meq/l) in seawater. Due to this difference and since all divalent species are rejected well by conventional NF membranes, Cl has to be also slightly rejected in order to maintain electro-neutrality in the product (retentate) solution. Nativ et al. [9] also showed, under such operation, that while the Cl concentration was maintained around 7.5 g/l (per Mg2+ concentration of ∼5 g/l) monovalent cations and uncharged species such as Na+, B and K+ were hardly rejected at all and their concentration at the end of the DiaNF step dropped to close to zero, as expected.

To overcome this physical inhibition (chloride ions retarded in the retentate during the DiaNF step to maintain solution electro-neutrality) a third filtration step was added in the current work: prior to the NF-DiaNF steps the seawater (or SWRO brine) was passed through a relatively “open” nanofiltration (or perhaps “tight” ultrafiltration) membrane characterized by dense negative surface charge at the target pH range (pH > 5). Such a membrane (in this work the polyethersulfone-based Microdyn-Nadir NP030 was used) is capable of rejecting SO42− better than Mg2+ and Ca2+ (due to the highly negatively charged surface) and does not reject monovalent ions at all. The result of this step is thus a retentate solution that is characterized by a TH to SO42− ratio that is lower than the original ratio in seawater or seawater reverse osmosis retentate (which is ∼2:1), the exact value depending on the recovery ratio (RR) applied in this step (RR = Qpermeate/(Qretentate + Qpermeate)) and the rejection characteristics of the membrane for (namely) SO42−, Mg2+ and Ca2+. The NF-DiaNF procedure is then applied on the retentate solution of this filtration step, now using a much denser NF membrane (e.g. DLF1021; GE). This membrane rejects all divalent ions well, but allows monovalent ions to permeate through it, resulting in a retentate solution with a much lower monovalent to divalent ions concentration ratio. The underlying hypothesis in this work was that reduction of the TH to SO42− ratio in the solution flowing into the NF-DiaNF treatment sequence would remove the inhibition posed on Cl passage through the NF membrane and thereby result in a lower Cl:Mg ratio in the product (retentate) solution, as required. In other words, the hypothesis is that chloride ions retain in the retentate is due to the lack of electro-neutrality of the divalent ions. To minimize costs the second NF step (Step II in Fig. 1) was designed to be performed at the highest recovery ratio possible before the permitted saturation index (SI = log(Q)  log(K′) where Q is the activities’ quotient of the precipitation equation and K′ is the apparent saturation constant, i.e. the thermodynamic saturation constant adjusted for temperature and ionic strength effects) of CaSO4 (in the presence of the applied antiscalant) was exceeded. H2SO4 was also dosed to prevent CaCO3 precipitation. Once generated, the Mg-rich retentate product of the DiaNF step is subjected to an ultrafiltration step with the purpose of removing microorganisms prior to be injected to the drinking water (results from this step are not shown in this paper).

A schematic of the suggested four-step process is shown in Fig. 1.

Section snippets

Experimental system

Both the NF and the DiaNF experiments were carried out in a pilot-scale desalination unit sustaining one 4″ high pressure vessel, an Osip riva-80 booster pump and a Hydra-Cell G10 high pressure pump. A 25 μm filter was used to protect the high pressure pump. A titanium heat exchanger and chiller were used to maintain a constant temperature of 24 ± 1 °C. All wetted piping was made of stainless steel or P.V.C. Flow rates and pressures were digitally and analogically measured. Two types of NF

Filtration step I

The aim of this step (in which the NP030 membrane was used), was to reduce the concentration ratio between total hardness (TH) (i.e. the molar sum of Mg2+, Ca2+ and Sr2+) and SO42− in the retentate that is treated in the following NF-DiaNF steps.

Table 2 shows the average species concentrations developing in the retentate of Step I experiments, as a function of the applied recovery ratio. It can be seen that a ratio range of 1.48–1.55 (M/M) was attained between total hardness and SO42− at 75%

Cost assessment

The estimated costs of most of the alternatives presented in this paper are shown in Table 8. The operating expenses (OPEX) were estimated by considering the costs associated with the following components: electricity demand in the two NF steps, the DiaNF and the UF steps, antiscalant dosage and the cost of the RO permeate to be used as diluting solution in the DiaNF step. The cost of erecting an NF plant with various modules (CAPEX) was adopted from the literature. The cost associated with the

Conclusions

  • A new three-step process (NF-NF-DiaNF) was introduced to selectively separate divalent ions from seawater in a cost effective fashion. The goal of the first NF step is to reduce the molar ratio between TH to SO42−. The 2nd NF step is used to both increase the Mg2+ concentration and the Mg2+ to monovalent ions ratio, thereby reducing the demand for costly diluting solution applied in the DiaNF step. The DiaNF step is applied to reduce the monovalent ions and uncharged species concentrations to a

Acknowledgement

The authors wish to acknowledge the contribution of the Uzi and Michal Halevi fund for innovative applied research at the Technion. The support of the Technion-Guangdong Fellowship to S.C.N. Tang is gratefully acknowledged.

References (22)

  • I.C. Karagiannis et al.

    Water desalination cost literature: review and assessment

    Desalination

    (2008)
  • Cited by (29)

    View all citing articles on Scopus
    View full text